Emotion Recognition for Student Learning Enhancement Using Multi-Layer Feature Fusion and AI Techniques

FER remains a focal point for research interests within computer vision due to its capability to enhance security measures for both human-computer interactions and healthcare system applications. Our work presents a method that combines features from various layers based on InceptionV3 CNN model guidance. The abbreviation for the Multi-Layer Feature-Fusion based Classification model is MLFFC. The MLFFC is unique as compared to other traditional models because it extracts features from different layers of network and in particular from Inception Module C. The novel feature-fusion inter-layer technique for the proposed model is also constructed in which the model fuses multiple network depths to learn face features from a much wider spectrum. Two benchmark datasets namely CK+ and FER2013 serve as testing grounds for evaluating the MLFFC model. The experimental results demonstrate that the MLFFC model reached the greatest possible accuracy ratings of 99.69% on CK+ and 70.29% on FER2013 which surpasses the performance of current methods. Face expression recognition models achieve enhanced stability and reliability for real-world applications by using combined features extracted from multiple network layers.

Ҳали таржима қилинмаган